![]() 194: Fix compute pass barriers r=grovesNL a=kvark Fixes #193 (supposedly) @m4b would you be able to confirm? Basic idea is that we treat commands inside a compute pass to follow the same rule as transfer commands. We could even go as far as alias `type ComputePassEncoderId = CommandEncoderId`, but I thought maybe we can still take advantage of a separate type. Currently it gets both the backend command buffer and the tracker from the primary encoder, only to return them back in place when the pass is done. Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com> |
||
---|---|---|
examples | ||
ffi | ||
wgpu-native | ||
wgpu-remote | ||
.clang-format | ||
.gitignore | ||
.travis.yml | ||
bors.toml | ||
Cargo.lock | ||
Cargo.toml | ||
CHANGELOG.md | ||
LICENSE | ||
Makefile | ||
README.md | ||
rustfmt.toml |
This is an active GitHub mirror of the WebGPU native implementation in Rust, which now lives in Mozilla-central. Issues and pull requests are accepted, but we merge them in m-c manually and then sync to GitHub instead of landing directly here.
WebGPU
This is an experimental WebGPU implementation as a native static library. It's written in Rust and is based on gfx-hal and Rendy libraries. The corresponding WebIDL specification can be found at gpuweb project.
The implementation consists of the following parts:
wgpu-native
- the native implementation of WebGPU as a C API librarywgpu-remote
- remoting layer to work with WebGPU across the process boundaryffi
- the C headers generated by cbindgen for both of the libraries
Supported platforms:
- Vulkan on Windows and Linux
- D3D12 and D3D11 on Windows
- Metal on macOS and iOS
Usage
This repository contains C-language examples that link to the native library targets and perform basic rendering and computation. To run the C triangle example, install a C compiler + glfw 3, then run these commands at the root of the repo:
rustup toolchain install nightly
cargo install cbindgen
make examples-native
cd examples/hello_triangle_c/build
./hello_triangle
The idiomatic Rust wrapper lives in https://github.com/gfx-rs/wgpu-rs and provides a number of more complex examples to get a feel of the API.